01
Open to opportunities · 2026

ZHANG JR

Ph.D. researcher turned AI engineer. Specializing in LLM/VLM post-training, Agentic RL, and building agents that plan, act, and learn in production.

4+
Years Exp
30+
Projects
8
Open Source
Scroll

About
Me


Ph.D. in Computer Science, specializing in post-training for large multimodal models. Currently leading algorithm R&D on VLM/LLM alignment, Agentic RL, and real-time video AI — shipping models that power production-grade intelligent agents.

I believe the next leap in AI is not bigger models, but models that can plan, act, and learn from interaction. My work lives at that frontier: teaching models to use tools, reason through uncertainty, and improve through reinforcement.

When I'm not running training jobs, I'm vibe coding side projects with AI pair programmers, reading new RL papers, or thinking about what it means to build agents that are actually useful.


2021~

Senior Algorithm Engineer / Tech Lead · KAOLAURAN Tech Corp

Led algorithm team on full post-training pipeline for large multimodal/language models (VLM/LLM), covering SFT, DPO, On-Policy Distillation, RLVR, and Agentic RL. Built distributed training infrastructure supporting 1.5B–70B+ models and a 15M-sample multimodal data pipeline. Drove deployment of 3 industry-specific LLMs (smart governance, safety production, tobacco), lifting recognition accuracy from 60% to 90%+. Established standardized model iteration workflows, automated multi-dimensional evaluation, and end-to-end training monitoring with 98% data quality pass rate.

2018-2022

Ph.D. Computer Science and Technology · University of Electronic Science and Technology of China

Research focus on multimodal video analysis, including audio-visual speech analysis, video action recognition, and adversarial attacks.

2015-2018

M.S. Communication and Information Systems · Jiangxi University of Science and Technology

Research focus on face recognition techniques including PCA, SVD, manifold learning, and deep neural networks.

2011-2015

B.Eng. Electronic Information Engineering · Wuhan University of Science and Technology

Core coursework in matrix analysis, statistical learning, data structures, and electronic signal analysis and processing.

Technical
Stack

🎯
Post-Training & Alignment
Full post-training pipeline from SFT to preference alignment and reinforcement learning. Experienced in on-policy and off-policy methods for both capability improvement and safety alignment.
SFTDPOPPOGRPORLVRRLHFReward Modeling
🏗️
Distributed Training
Building and maintaining large-scale distributed training infrastructure across multi-node GPU clusters, with focus on throughput optimization and training stability.
DeepSpeedMegatron-LMFSDPRayNCCLFlashAttention
👁️
Multimodal & VLM
End-to-end experience with vision-language models covering architecture adaptation, multimodal instruction tuning, and video understanding tasks.
VLMViTVideo LLMMultimodal SFTVisual Grounding
🗂️
Data Engineering
Designing large-scale multimodal data pipelines including collection, filtering, annotation, and synthetic data generation for post-training.
Data PipelineSynthetic DataQuality FilteringAuto-AnnotationParquet/HDF5
📊
Evaluation & Benchmarking
Building automated multi-dimensional evaluation systems for LLM/VLM, covering capability benchmarks, reward hacking detection, and production monitoring.
Eval HarnessLLM-as-JudgeMMMUMT-BenchCustom Metrics
⚙️
MLOps & Inference
Deploying and serving large models in production with focus on inference acceleration, cost efficiency, and training-to-serving pipeline automation.
vLLMTGIQuantizationLoRA ServingK8sTriton

Selected
Work

001

Video Agent Agentic RL Training System

Replaced rule-based workflow engine with a ReAct-architecture VLM Agent capable of autonomously orchestrating 70+ atomic CV tools. Designed hierarchical action space and multi-dimensional verifiable reward function. Built fully async Inference-Training engine on verl, lifting GPU utilization from 60% to 85%. Composite task completion rate reached 85%+.

Agentic RLGRPOSAMPOReActverlPRM
002

Multimodal Reasoning Model

Reproduced and optimized DeepSeek-R1 RLVR framework with a hybrid RL+DPO training strategy. Introduced dynamic importance sampling and clip-higher policy to stabilize training. Implemented On-Policy Distillation enabling cloud-to-edge model evolution.

RLVRDPOOn-Policy DistillationQwen3-VLMoE
003

Real-Time Multimodal Video Interaction System

Designed a low-latency audio-video interaction system supporting visual understanding and external tool calls beyond traditional audio-only voice assistants. Async event-driven pipeline with 20+ processors integrated 15 AI service providers (OpenAI, Google, Anthropic, etc.). Achieved <500ms end-to-end latency supporting 8-channel 1024p@30fps concurrent streams.

VLMVADStreaming STT/TTSAsync PipelineWebRTC
004

Multimodal Training & Data Infrastructure

Built dual-track SFT system for Dense (1.5B–70B) and MoE (~30B-A3B) models on ms-swift, raising GPU utilization to 90%+. Designed modular VLM framework with flexible vision encoder and alignment layer combinations. Constructed end-to-end data pipeline producing 15M multimodal instruction samples at 98% quality pass rate.

DeepSpeedMegatronms-swiftSequence Packingverl
005

Converge — Kubernetes-Inspired Agent Framework

A production-ready Python framework for orchestrating AI agents at scale. Declarative API maps Kubernetes concepts (Deployment→AgentWorkflow, kubelet→AgentRuntime, etcd→StateStore) to agent orchestration — declare the goal, the ReconcileLoop converges to it. Supports DAG-based multi-agent workflows, RBAC tool permissions, Human-in-the-loop for high-risk actions, and full OpenTelemetry observability.

AgenticReconcileLoopLiteLLMasyncioPydantic v2PostgreSQL

Let's
build
together.